Thanks for all the replies, and after having read back over my original post I feel I must apologise for it's bloated length :) <<<
I for one don't have a problem; I think you're making a lot of very good points. So carry on.
However, this does pose some problems. The decentralised peer review process works fantastically to a point, but when a product needs something that is outside the requirements of the peers themselves - it becomes unlikely for it to happen. Because the developers are perfectly capable of using a command line interface, they're unlikely to claim a fault in an application being that it has no intuitive GUI that could be picked up and used by anyone.
I suspect that given the huge number of people out there, all with differing needs, pretty everything that's needed gets written. The problem isn't with the code; it's usually with the documentation. Geeks (on the whole) aren't the most literary-minded people and they don't (again on the whole) much like writing manuals for non-geeks. Many years ago I dicovered that Unix documentation was very useful as a reference but not much help as a user guide. Once I'd been through the pain of discovery, most of the documentation could be summarised as
"So that's what it meant!"
And not a lot has changed. This is as true for Windows as for Linux; does the Windows on-line help really tell you from scratch how to install a network card? Don't think so.
If Linux is to really beat Windows to the totally-non-geek market we need a lot more literate types to get down in the software and describe to the rest of us how it all works. The Linux Documentation Project is good - very good in places - but I feel there's still a long way to go.
-- GT